As the complexity of logistics scenarios continues to increase, numerous new challenges have emerged. In multi-depots pickup and delivery path planning (MDPDP) scenario, these challenges include cross-depot resource coordination, spatio-temporal dependencies of pickup-delivery pairs, and generalization across problem scales. To address these challenges, this paper proposes a deep reinforcement learning framework with a Heterogeneous Collaborative Multi-Attention (HCMA) mechanism. The HCMA framework has three key features. 1) HCMA comprises a three-level heterogeneous attention architecture: an inter-depot collaborative attention module for cross-depot resource allocation, a pickup-delivery pairing attention module that explicitly learns sequential and pairwise relationships between task nodes, and a global attention module to capture long-range node dependencies. 2) HCMA dynamically integrates real-time vehicle status with multi-type node features through attention weighting to balance vehicle resource utilization across depots. 3) HCMA adopts multi-start training and instance augmentation strategies to enhance generalization capability and obtain the global optimal solution. Extensive experiments conducted on random datasets, benchmark datasets, and a real-world Shenzhen logistics dataset demonstrate that the proposed HCMA method outperforms state-of-the-art heuristic algorithms and deep reinforcement learning baselines, while exhibiting superior generalization capability. The source code is publicly available at:https://github.com/CIA-SZU/LAQ

The memory dirty page prediction technology can effectively predict whether a memory page will be modified (dirty) at the next moment, and is widely used in virtual machine migration, container migration and other fields. In this paper, we propose a machine-learning based method for memory dirty page prediction. The method exploits the temporal and spatial locality principle of memory changes, collects dirty records of pages over a period of time, and uses supervised learning methods for training and predicting the dirty page. We also discuss the influence of data contradiction and data repetition in memory page dataset. The experiments with different memory change frequency dataset show that compared with the traditional time series methods, our machine-learning based method has better performance.

Clustering is a task of unsupervised learning, aiming to group a set of data so that data in the same group are more similar to each other than to those in other groups. Affinity propagation (AP) is a clustering algorithm which finds the exemplars (representative points) for data points by spreading messages among them. AP algorithm has several drawbacks. First, it is time-consuming and memory-consuming for clustering on large-scale dataset, due to its N square time and space complexity. Second, AP may produce too many small clusters. Third, AP may have difficulty in converging which leads to a higher cost of time for fine turning. To achieve better effectiveness and efficiency, in this paper we propose Mini-Batch Affinity Propagation (MBAP). MBAP processes small batches of data serially and obtains clustering results gradually. We also proposes MBAP with early stopping (MBAP_ES), which integrates MBAP with stopping strategy so that it can stop clustering early when the model is nearly unchanged. The experiments show the effectiveness and efficiency of MBAP and MBAP_ES in comparison to other AP-based algorithms.

Container live migration can move the running container between different physical machines without interrupting the operations of container, which brings great flexibility to cloud computing environments. The pre-copy algorithm is a commonly used migration algorithm in live migration. Aiming at the problem of repeated transmission of dirty pages in the pre-copy algorithm, this paper proposes a container pre-copy migration method called MBDPC, which based on dirty page prediction and page compression. The MBDPC method exploits the locality principle of program operations, uses the Random Forest model for dirty page prediction, and skips the transmission of repeated dirty pages. The MBDPC also uses the page incremental compression method in the transmission process, which further reduces the amount of data transmitted and the migration time. Experimental results show that the MBDPC method can significantly reduce the time and data transmission overhead of live migration compared with traditional pre-copy method.

Portfolio is a financial concept to combine several stocks to reduce the risks and improve the profits. To choose the basic members of portfolio, we can group similar stocks into one cluster and then choose representative stock from each cluster. In this paper, we focus on the method of choosing representative stocks in clusters. The ordinary representative of a cluster is often the center of that cluster. We propose a new cluster representative method MDR (maximum distance representatives). In our method MDR, we choose the stocks which has maximum distance with other representatives. MDR can construct a more diverse portfolio than center method. The effectiveness of cluster representative selection methods can be evaluated by an index IBEF based on the concept of efficient frontier. Our experiments show that MDR can effectively improve the efficient frontier, which means MDR can bring more profits than center representative method at the same risk level.

Clustering stocks by their time series data is a significant but challenging task in computer supported financial decision systems. In this paper, we propose an effective stocks clustering method based on hybrid correlation coefficient called SLU correlation coefficient which is a weighted combination of Spearman rank correlation, upper tail correlation and lower tail correlation. The upper and lower tail correlation is defined by Copula function and estimates parameters by EM algorithm. The similarity matrix is defined by SLU and inputs into Affinity Propagation algorithm for clustering. The experiment shows the effectiveness of the SLU, compared to Pearson correlation and DTW distance.

The similarity measurements of business processes have important applications in business process management, such as process model search indexing, facilitate reuse, processes merge, etc. Existing researches are mostly based on the syntactic or semantic of text labels of activities or tasks. This paper presents a method for similarity measurement based on the internal structure of business processes without considering the text labels of process activities. Petri nets are used to define the process models. The nodes (transitions and places) are mapped by an iterative mapping strategy to identify the correspondence of two Petri nets. After getting a stable best mapping of places and transitions, we can compute the similarity mesurement of two processes. Experiments on the real data sets show that our algorithm is reliable and effective to the actual demand.

The fast evolving workflow technologies facilitate organizations to interact and cooperate with each other to achieve their business goals by process collaborations. Task-role based access control is an important security mechanism to protect data and resources in information systems. However, the traditional centralized authorization and administration mechanism in access control can not satisfy the administrative requirements in process collaboration environments. In this paper, we propose a domain based administration model for task-role based access control (DATRBAC), in which the authorization and administration permissions are distributed to multiple administrative domains and administrative roles. Then we propose the solution to detect and resolve the conflicts between access control policies defined by different administrative roles. We also described the implementation of the model in the PLM product and the experiments based on the practical application data.

Role-based access control (RBAC) models have been successfully implemented in various information systems in recent years. However, the traditional centralized authorization and administration mechanisms in RBAC have several drawbacks in collaborative environments. In this paper, we propose a distributed domain administration of RBAC model, DARBAC, in which the authorization and administration privileges are distributed to multiple administrative domains. Each administrative role is assigned to an administrative domain and can only execute administrative operations within its domain. By introducing the concept of administrative domain and administrative role hierarchy, the DARBAC model can flexibly meet the access control requirements in collaborative environments. We also describe how to implement the model in the PLM product and how to apply the model in a distributed enterprise environment to support cooperative work